The Sui Move has emerged as a next-generation smart contract language solution, emphasizing scalability, low latency, security, and robustness. Sui Move is built around a unique and innovative object-centric model, which introduces a new paradigm of flexibility in asset management. The Sui Move object-centric data storage model is particularly advantageous for managing complex assets, offering a more efficient and secure way to interact with other on-chain objects. Due to its high-performance and scalable architecture, Sui has emerged as one of the fastest growing Layer 1 blockchains in industry, particularly in the DeFi and gaming sectors. In this paper, we present the first analytical study on the performances of Sui object-centric models. We will describe the Sui gas pricing mechanism. We will discuss the rich Sui Move object-centric models, including wrapped objects, dynamic fields, dynamic object fields, and dynamic collections. We will comprehensively study the fees, object hierarchies, and object accessibilities in these models in both small scale and large scale. We will also provide sample smart contract code to implement these models. This is the first paper that provides a comprehensive analysis on Sui Move object-centric models and recommendations for Sui developers to develop efficient and cost-aware smart contracts. Our results indicated that developers could save significant fees by selecting an appropriate model for their smart contracts.

A Network Intrusion Detection System (NIDS) is a technology to monitor the network traffic and detect and classify the possible cyber threats towards a computer network. In case it detects a potential cyberattack, it will alert the network prevent and defeat security system. Identifying the type of attack is invaluable for such system to block the malicious traffic or mitigate its impacts on the network. The Internet of Things (IoT) is a system of devices connected to the Internet, mainly for the purpose of acquiring data and transmitting that to the cloud, where the data will be processed, analyzed, and stored on the cloud and used by the user. In many cases, a group of IoT devices located at the same place form a network and send their data to the cloud through the network’s gateway. As IoT technology is rapidly growing, it is crucial to develop more accurate and efficient intrusion detection and classification systems. In this paper, we propose efficient deep learning models that classify the network traffic to benign vs seven main types of IoT attacks with a focus on CICIoT2023 dataset at the edge of IoT network by leveraging edge computing. This research addresses the challenges of applying deep learning on intrusion detection datasets, including data imbalance and data distribution discrepancy using data preprocessing techniques, such as random under sampling, class weighting, and feature scaling. We also study the impact of different feature selection techniques on the efficiencies of our models. Our experimental results showed that our 1D-CNN model outperforms the state-of-the-art with 93.8% F1-score.

Over the past few years, the use of emails and text messages has drastically increased. Short Message Service (SMS) on cellphone providers and related apps, like Whatsapp, is one of the best and fastest ways to communicate among users. SMSs are used and sent globally for personal and business purposes. However, alongside safe SMSs, the users may receive fraudulent Spam SMSs, which could cause security issues and inconvenient for the users. Numerous Spam messages are being sent daily for both personal and professional benefits. Accurately identifying Spam SMS is a challenge. The objective of this research is to build a model utilizing machine learning and deep learning to understand the semantics of SMSs and classify them to either Spam or non-Spam (Ham). We used a pre-trained BERT model and combined it with several machine learning and deep learning models. The results indicated that BERT+SVC and BERT+BiLSTM performed the best with 99.10% and 99.19% accuracies respectively on the test dataset.

The rise of ransomware has emerged as a pressing concern for the technology industry, demanding prompt action to prevent monetary and ethical exploitation. Therefore, an accurate approach is imperative to identify and thwart such attacks effectively. Most of the prior ransomware detection techniques either are signature-based, which are inefficient to identify new ransomware, or utilize a dynamic analysis, which are complicated and computationally expensive. This paper proposes a feature selection-based framework along with different machine learning and deep learning algorithms that can effectively detect ransomware based on features extracted from the files. We performed various experiments beginning with filter, wrapper and embedded methods of feature selection and then applied Decision Tree (DT), Random Forest (RF), Naïve Bayes (NB), Logistic Regression (LR), Support Vector Machine (SVM), k-Nearest Neighbor (KNN), Extreme Gradient Boost (XGB) and Multi-layer Perceptron (MLP) on a ransomware dataset that contains the features and label from files. The experimental results demonstrate that RF and MLP classifiers with ANOVA filter method of feature selection outperform other methods in terms of accuracy, precision, and recall.

The scale and scope of using IoT has had a rapid growth over the past few years. This growth has raised security and cost challenges of using IoT smart devices in large scale. In this research, we propose a comprehensive framework that addresses these challenges using cryptographic systems and blockchain NFTs. We analyze the credibility and scalability of our framework. In our proposed framework, we securely authenticate IoT devices and bind them to blockchain Non-Fungible Tokens (NFTs). To uniquely bind each IoT device to an NFT, we use the device's physical unclonable function (PUF). The link between the NFT and the device is difficult to break and can be traced anytime. Our framework also authenticates the device's acquired data using the device's PUF and securely stores the authenticated data on the blockchain. We use an approach that significantly reduces the blockchain costs and analyze it in large scale. Our analysis show that our framework is a secure and cost-effective solution for large-scale IoT authentication and data storage.

The rise of network attacks has emerged as a pressing concern for companies and individuals. A Network Intrusion Detection System (NIDS) is employed at the network edge and monitors the traffic exchanged between the network and the cloud. An anomaly detection NIDS detects benign vs attack traffic while a misuse detection NIDS detects benign vs. specific attack. In this paper, we propose two machine learning based NIDS frameworks, an anomaly detection, and a misuse detection NIDS that can effectively detect and classify IoT network attacks. We used Kitsune IoT network attack dataset that contains 9 types of attacks. We used random undersmapling to reduce the dataset size. Then, we employed ANOVA and Chi-square feature selection techniques and studied a variety of machine learning algorithms for binary classification (anomaly detection) and family classification (misuse detection). We studied different models to find the best combination of feature selection method, number of features, and hyperparameters for each model for both binary and family classifications. Our experimental results demonstrated that our proposed Random Forest and Extreme Gradient Boost binary classifiers and Random Forest family classifier with ANOVA feature selection outperform other models and existing research in accuracy.

An intrusion detection system is a technology built to monitor, detect, and prevent malicious activities and cyberattacks towards computer networks. The Internet of Things (IoT) is a system of devices connected to each other and to the Internet that facilitates communication between the IoT devices and the cloud. This rapidly growing technology calls for more accurate and efficient cyberattack detection techniques to ensure the security of IoT devices and systems. In this paper, we focus on using machine learning and deep learning algorithms along with feature selection methods to detect cyberattacks effectively at the edge of IoT network by leveraging multi-access edge computing. This study addresses the challenges of processing intrusion detection datasets, such as data imbalance and missing data, with a focus on UNSW-NB15 dataset. We use ANOVA and embedded feature selection techniques and apply various machine learning algorithms, consisting of Decision Tree (DT), Random Forest (RF), Light gradient-boosting machine (LightGBM), Artificial Neural Network (ANN), K-Nearest Neighbor (kNN), and Extreme Gradient Boost (XGB), on UNSW-NB15 dataset. Our experimental results indicate that our classifiers achieved higher accuracies and efficiencies comparing to state-of-the-art machine learning intrusion detection approaches and our LightGBM model is the most accurate and efficient one among all.

The size and scope of using IoT has been rapidly growing in the past few years. This growth rises security challenges in networks. One of the pressing concerns is detecting the type of attacks emanated from IoT devices. To tackle this issue, machine learning solutions have been proposed that classify the IoT traffic. Most of the current solutions on IoT botnet attack family classification propose a separate model for each IoT device. This is not a suitable approach for an IoT ecosystem where a variety of IoT devices are used and new device types are introduced every day. In this research, we propose a united model that classifies the traffic from any type of IoT device. Such approach is especially essential in IoT as a large number of different devices could be infected to be used in large-scale botnet attacks. We propose CNN-based deep learning family classification models, which classify the IoT traffic to benign and different types of botnet attacks. We trained and tested our models using N-BaIoT dataset, which contains data for benign traffic and 10 types of BashLite and Mirai botnet attacks. The experimental results demonstrate that our models outperform the related proposed models in terms of accuracy, precision, and recall.

The fast-growing Internet traffic is increasingly becoming content-based and driven by mobile users, with users more interested in data rather than its source. This has precipitated the need for an information-centric Internet architecture. Research in information-centric networks (ICNs) have resulted in novel architectures, e.g., CCN/NDN, DONA, and PSIRP/PURSUIT; all agree on named data based addressing and pervasive caching as integral design components. With network-wide content caching, enforcement of content access control policies become non-trivial. Each caching node in the network needs to enforce access control policies with the help of the content provider. This becomes inefficient and prone to unbounded latencies especially during provider outages. In this paper, we propose an efficient access control framework for ICN, which allows legitimate users to access and use the cached content directly, and does not require verification/authentication by an online provider authentication server or the content serving router. This framework would help reduce the impact of system down-time from server outages and reduce delivery latency by leveraging caching while guaranteeing access only to legitimate users. Experimental/simulation results demonstrate the suitability of this scheme for all users, but particularly for mobile users, especially in terms of the security and latency overheads.

Wireless ad hoc networks (WAHNs) consist of autonomous nodes cooperating with each other to transmit/receive data over multiple-hops in the network. Caching is a useful mechanism to leverage this cooperation. Nodes with cached content can satisfy requests from other nodes, thus helping reduce network traffic and energy consumption, and improve latency. With the proliferation of wireless devices on the Internet and the proposal of a future Internet with emphasis on in-network caching, improvements in caching can significantly improve network response while reducing network load. In this paper, we present a holistic caching framework, Split-Cache, which enables a network node to account for the frequency of requests of data items and their presence in the network, and to leverage a split-cache (one part caches popular items and the other caches less popular items) to make caching and cache-eviction decisions. We performed exhaustive simulations to compare Split-Cache with the state-of-the-art: Split-Cache improved the cache request resolution time on an average by 30% (and as high as 72%), and required 15% less average traffic for resolving requests-large savings when considering large number of requests.

The constrained relay node placement problem in a wireless sensor network seeks the deployment of a minimum number of relay nodes (RNs) in a set of candidate locations in the network to satisfy specific requirements, such as connectivity or survivability. In this paper, we study the constrained relay node placement problem in an energy-harvesting network in which the energy harvesting potential of the candidate locations are known a priori. Our aim is to place a minimum number of relay nodes, to achieve connectivity or survivability, while ensuring that the relay nodes harvest large amounts of ambient energy. We present the connectivity and survivability problems, discuss their NP-hardness, and propose polynomial time${\mbi{\cal O}}$(1)-approximation algorithms with low approximation ratios to solve them. We validate the effectiveness of our algorithms through numerical results to show that the RNs placed by our algorithms harvest 50% more energy on average than those placed by the algorithms unaware of energy harvesting. We also develop a unified-mixed integer linear program (MILP)-based formulation to compute a lower bound of the optimal solution for minimum relay node placement and demonstrate that the results of our proposed algorithms were on average within 1.5 times of the optimal.

The constrained relay node placement problem In a wireless sensor network is concerned with deploying a minimum number of relay nodes (RNs) in a set of candidate locations in the network to satisfy a specific requirement(s), such as connectivity or survivability. In this paper, we study the constrained relay node placement problem in an energy harvesting network. In such a network, it is imperative that the placement be energy harvesting aware, because the more energy the placed nodes can harvest the more effective the network can be. In our study, the RNs are constrained to be placed at only the candidate locations, where the energy harvesting potential of the locations are known a priori. Our aim is to place a minimum number of relay nodes, to achieve connectivity or survivability, while ensuring that the relay nodes harvest large amounts of ambient energy. For both the connectivity and survivability, we study the problems, prove that they are NP-hard, and propose polynomial time O(1) approximation algorithms with low approximation ratios. We also validate the effectiveness and efficiency of our algorithms through simulations and show that the RNs placed by our algorithms harvest 50% more energy on average, in comparison to those placed by the algorithms unaware of energy harvesting.